• Login
    View Item 
    •   ResearchSpace Home
    • College of Agriculture, Engineering and Science
    • School of Agricultural, Earth and Environmental Sciences
    • Geography
    • Masters Degrees (Geography)
    • View Item
    •   ResearchSpace Home
    • College of Agriculture, Engineering and Science
    • School of Agricultural, Earth and Environmental Sciences
    • Geography
    • Masters Degrees (Geography)
    • View Item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Refining prediction accuracy for pest blackfly outbreaks using Bayesian networks, Orange River, Northern Cape, South Africa.

    Thumbnail
    View/Open
    Naidoo_Sashin_2018.pdf (4.691Mb)
    Date
    2018
    Author
    Naidoo, Sashin.
    Metadata
    Show full item record
    Abstract
    Construction of dams, impoundments and Inter-Basin Transfer schemes (IBTs) along the Orange River are aimed to provide useable water for multiple sectors. However, operation of these water schemes had led to changes in flow regimes, seston concentrations and water temperatures, which has led to an escalation of pest blackfly (S.chutteri) outbreaks along the lower to middle reaches of the Orange River. Pest blackfly bite livestock, poultry and humans asthey require a blood meal to complete ovarian development. During outbreak periods, livestock farming and the grape industries are affected negatively by pest blackflies along the Orange River. The blackfly control programme has been operating for over twenty years, and aims to control blackfly outbreaks by applying larvicides along the Orange River. Although this programme is in place, periodic outbreaks occur and losses in livestock and productivity can amount to an estimated R300 million during an outbreak (2013). Therefore, other methods should need to be integrated with this programme to achieve blackfly control. Predictive modelling was identified as a method to assist the blackfly problem. Being able to predict when, where and the severity of an outbreak, will assist management in control planning. Bayesian network (Bn) models were identified as a suitable predictive model,as multiple variables can be used in understanding the cause and effects of a response variable.The aim of the research was to refine prediction accuracy of blackfly outbreaks along the middle to lower reaches of the Orange River, using Bns. Fourteen sites were sampled along the Orange River, for which abiotic and biotic data were collected during four sampling periods. These data were used in assisting quantitative components of the Bns, whilst the qualitative components were based of previous Bns with additions on new nodes that were identified as affecting blackfly outbreaks. Water temperature data showed that sites were split into two distinct groupings, for which Bns were constructed.These were termed the upper and lower stream models. The upper stream model had the higher outbreak probabilities, whilst it was predicted for both models that summer would be the season most likely for an outbreak to occur. The species most likely to cause an outbreak was identified to be either S.chutteri or S.damnosum, with switching in dominance throughout sampling periods potentially due to switching in seston concentrations. Future outbreak probabilities based on scenarios of increased discharge and water temperatures indicate that the blackfly problem is likely to worsen, with increases in discharge resulting in greater habitat availability for pest species and increases in water temperature resulting in shorter life cycles and more rapid reproduction.The Bns constructed show promise in assisting management as blackfly outbreak probabilities were refined on a spatial and temporal scale along the middle to lower reaches of the Orange River.
    URI
    https://researchspace.ukzn.ac.za/handle/10413/16542
    Collections
    • Masters Degrees (Geography) [206]

    DSpace software copyright © 2002-2013  Duraspace
    Contact Us | Send Feedback
    Theme by 
    @mire NV
     

     

    Browse

    All of ResearchSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsAdvisorsTypeThis CollectionBy Issue DateAuthorsTitlesSubjectsAdvisorsType

    My Account

    LoginRegister

    DSpace software copyright © 2002-2013  Duraspace
    Contact Us | Send Feedback
    Theme by 
    @mire NV